Discussion on Session 1: Introduction to Ethical issues in Cluster Randomized Trials

Question/Comment from the audience: I’m Diana Elbourne from the London School of Hygiene and Tropical Medicine in London, England. Happy to be here. Well I have three questions but I know I have other opportunities to ask them, so I try not to be too greedy. Allan, when you talked about consult statement right at the end, I just, this is a point of information really, just to say it's being updated actually as we speak, it's about to be submitted for publication and thanks to Jeremy and Monica, we are actually bringing in something about ethics which isn't currently there in the previous drafts and that's because all the consult statements work on the principle that ethics is a given, that informed consent is a given and we don't specially ask for that information to be reported and that might be reflected in the empirical work that you've been doing that doesn't show that. So, based on your pressure and pleading, we're going to add this in the next draft. The second thing is, just a point that always irritates me, which is that you're always talking about research subjects and the way we all trying and work is to consider the people that are taking part in our trials not to be subjects but to be participants in the research endeavor and so, I know you've got lots of publications about that already have that in the title, but perhaps not to do that in the future. And then the last thing is really a question, those are all points. Very often, you talk about things that you are trying to contrast, just in trials with individually randomized trials and all of you have been doing that. But a lot of the points that you make, trying with me, who does a lot of work in emergency trials, as being not that different where it's actually quite difficult to get consent in a practical sense. It may not be ethical to be trying to ask those questions and so, you often have proxies instead of gatekeepers and so lots of the issue may be rather similar and not as distinct as perhaps we are trying to make here, so that's a rather controversial point to bring in.

Question to ask? Jeremy Grimshaw: Just to say that, uh, I think Diana is right in terms of there are some similarities with particularly trial emergency settings. I guess the issue is the cluster controls trials in particular sort of family trials where these things are always problematic and there are an increasing number of cluster randomized controlled trials and so the feeling of trying to understand what the ethical issues of cluster randomized trials are. That could help the discourse between trialists and the research ethic boards that might be useful but certainly we like to learn things and our findings might be rather that we find in other types of studies, so I think it's a good point.

Question/Comment from the audience: I'm Kathleen Lohr from RTI International in North Carolina, USA. This is a question probably for Jeremy but perhaps all three of you might care to respond. I haven't heard yet anything about the complications that might arise if in fact if you were to pay any kind of incentives or honorary or whatever you might want to call it, to either the people who are in the clusters, let's say the GP's or whatever, or possibly to the people to whom you may care to collect data, which would be our research participants, our patients, our whatever, and I'm wondering if you've considered that as a peculiar aspect that complicates even more the design or what might happen in respect to the ethics issues when you're paying some people something and perhaps, not paying others that you might well should have.

Jeremy Grimshaw: Again, I think it's a very interesting comment. It, when we did our initial scan, it wasn't an issue that we identified as being particularly pointed for cluster randomized trials and as you were talking, Kathy, until your last sentence, I was thinking, "Well, this isn't particular to cluster randomized controlled trials." Your last sentence, which actually pointed out that you, may end up or maybe if you’re like an inequitable distribution of incentives, because of the structure of cluster randomized controlled trials, maybe we need to think about it. I think what will happen is, we may have some discussion today and over the next few days but we're going to, at some point, identify new ideas that require some thought. Think about them, cause when you look at the six questions, they've been subject to basically 3 years of fairly intensive thought and I don't think anybody would want to give an immediate response but I think we will take that away and think about it.

Question/Comments from the webcast: So, the first question from the webcast comes from Emma Veitch and the question is along the lines of. The presentations outlined some major hurdles to overcome in identifying ethical issues in cluster randomized trials and the question relates the point raised by the first live question here in Ottawa is: Irrespective generating guidelines for ethics committees on how to address these issues, do you think there's a need to further encourage trialists to report in greater detail how ethical issues were considered in the conduct of their trials? Monica Taljaard: Yes, I think so; Diana mentioned with the next integration of the consort extension of the cluster randomized trials, we will actually be adding some recommendations regarding ethics reporting.

Question/Comments from the webcast: Question from Michel Wensing in the Netherlands. Triggered by some major recent scandals and sites in my country where researchers had completely faked data, I wondered whether there was anything specific to cluster randomized trials along the lines of addressing scientific integrity issues that need to be addressed and whether there's something specific to cluster randomized trials or not that we need to consider? Jeremy Grimshaw: Hi Michel, thank you for the question. My major response, I'm again trying to think about what is peculiar to cluster randomized controlled trials, so the whole issue of research integrity is clearly a major issue across all of science. We have lots of examples where current mechanisms have failed to I think, ensure integrity. But my main review, I can't think of any obvious things in cluster trials that are different from patient trials or other ideas. If either you have and you want to come back that would be great or if other people in the room have, that would be really good but I can't think of anything that would immediately say this is unique to cluster trials as opposed to, this is something that is problematic across all the science that we do. Allan Donner: Well, perhaps issues regarding the choice of ... I'm just saying, maybe issues relating to the appropriateness of a particular gatekeeper in a cluster trial. Is it appropriate, for example, for a mayor of a city to randomize their residences? Did they elect their mayor to do that? I'm not sure, my mayor tried to widen my street by 8 feet the other year without asking my consent but that may be real.

Question/Comment from the audience:
Hi, I'm Sarah Edwards from London, UK. I throughout enjoyed all of your presentations and I've got umpteen things going around in my head but for the sake of everyone's patience, I'll limit myself to one question and one brief comment if I may. The question is for Monica. Did you attempt to do any analysis of interactions between the responses? So, for example, I was thinking in particular, the extent to which there might have been an association between people identifying research subjects and their opinions on the need for consent, whether they went hand and hand. Monica Taljaard: I expect that is likely the direction we will go because as you'll see later this afternoon, a little bit later, we do actually find an interaction between who we would consider to be research subjects and the need for informed consent. I won't give that away, but we're at the very beginning stages of analyzing that data so that will be something interesting to look at. Thanks. Sarah Edwards: And my brief comment is for Diana. Just on the whole business of the words we use to describe people in research, research subjects or research participants. I wondered to what extent we should be looking at how involved individuals are before labeling them participants? I think, I mean, there will be some cases in cluster designs where the individuals aren't obviously doing anything, they may not even know that they are part of a study so to call them participants might be a bit misleading.

Question/Comment from the audience: Roger Soll, from the University of Vermont. I assume that all cluster trials that are being conducted today are registered and I don't know what information is available in terms of registration websites and I guess my question is two-fold. One, have you looked at registration websites to see if people are describing more of these important features currently or even if this type of information is asked for by their websites? So that if some, as part of your consort statements, since these things have to be planned prospectively, they really should be wrapped into. I don't mean to make more work for your group, but it should be wrapped into that and perhaps journals are always loathed to publish some of the methodology. That's the place where a lot of this rich and important information could be shelled, so what do we know about registration sites? Jeremy Grimshaw: So registered trials, clustered trials on registration sites, I honestly, I have not looked at the field to the extent to which they represent sort of unique challenges in cluster trials. It would be worthwhile following up with that. I suspect cluster trials will register their trials for exactly the same reason that individual patient trials will be because of the other requirement from fund lenders or regulators to do that, so I don’t think there's a problem with registration. Whether it is fully informative for cluster trials is less clear and I do know the studies sort of try to look at the quality of data in trial registers found that they are pretty poor. They have missing data fields or inaccurately completed, so there may be an issue about guidance or more structured reporting within trial registered per say. The final thing also is that I think what we probably need to do is move forward in making trial protocols more readily available. They may be something that can be posted as part of a trial registration process or to put a plug in trials which is a bi-centennial open access journal which I am Editor in Chief on and journals like Inflammation Science which I am also Editor and Chief on, routinely publishes study protocol to allow the trialists to have length to actually describe what they've done. Allan Donner: Yeah, I don't think there's been much guidance on what elements of a cluster trial to put in a registry so your point is very well taken. For example, just with respect to sample size, you could specify the number of clusters or you can specify the number of individuals per cluster or both so I think that's a very good point. Roger Soll: My guess is that it's a square peg and round hole thing and may actually be steering investigators in the wrong direction as they think about all the sample size issues and whose being randomized and the ethical issues and so I think that changing people's thoughts around registration might be very productive but actually I take Jeremy’s point about getting the whole protocol out there is probably where it's going to be.

Question/Comment from the audience:
Hi, good morning. I'm Dr. Abha Saxena. I'm from WHO in Geneva. I have a couple of comments and perhaps one question. First comment was I agree with the first speaker on some of these guidelines that you're developing, not necessarily only for cluster randomized trials. They apply very much to other types of research studies. For example, we've been doing work on patient safety research, but also held system research more broadly other than just the cluster randomized part of the health system. So at some point in time, I think we'll have to say, ok, let's look at this under one umbrella, but perhaps not the time. I was, the second comment I have before I come to my question is related to the point about how many articles reported, whether they were reviewed by ethics committees or informed consent was taken, etc. And at one level, I think a lot of the people who are doing these types of studies, I think there's a disconnect in their mind for at least for some of these studies, the disconnect being that this is not real research, this is, we are just trying to see how practice, we are trying to think of practice, so it's not really research, so perhaps all these things don't work for us. We don't need to do all this but then because we are using research methods and we want to publish, so there is that disconnect and I think somewhere along, if we can have a better clarification for researchers and for others to say, what is research, what is not, that might be useful. Doesn't mean if it's not research it doesn't have implications but that the other things that kick in, if it's research. And third, my question was, you said, who should we consider the research subject? And I was just wondering, when you talk about who should be our research subject or participant, when you say who should be considered a research participant, is it, are you thinking in terms of, so that informed consent should be taken? So is that a research participant equals consent? Is that the issue? Or is it the issue that research participant equals participation? So, just clarity on what is meant by this particularl question would be useful. Monica Taljaard: For us, I believe the implications are, for a research subject, that it would be somebody who would need the protection of the ethics review committee and not necessarily implying that we need to seek informed consent. There will actually be a detailed presentation about that in the next session.

Question/Comment from the audience: Hi, I was vaccinated last week, so I don't worry about closeness of microphone. My name is Karmela Krleza-Jeric, CIHR, which is a Canadian Federal Government Agency. It's nice to come after other people so you can comment on what they commented. As far as the registration goes, CHI's funding what I now learned is cluster trials cause you are funding both randomized trials which are clinical and others ones are groups of people. And according to regulators, I don't know any regulator who is asking registration of those trials. Usually regulators are looking at new drug and the testing of the new drug. So seldom customer trials who go there. So that's the issue saying we won't publish unless. So that's the comment about the registration. And then, my question actually was, I've been before I came to Canada, I've been involved in what we called interventions. We didn't even think it was research. I worked in public health and we were doing smoking cessation or whatever and followed up and I think some of them even got published. That was the beginning of my career but, so my concern here is that we should probably look into different cluster trials according to whether you test a new intervention or you are applying the known, like smoking cessation is not a new discovery. Sereca Dolce did it before 50 years ago, almost. So in that case, let's turn it around the table and say how dangerous is, potentially dangerous is, intervention to lower participants, so to say, another cluster. Let's say we talk to GP's and they do something different with their people we talk to. So I think we should distinguish those two types of cluster trials we are talking about consent form and everything. And I'll give you an example of the study we did, we randomized two groups of hospitals. A doctors and in Yugoslavia in those days, everybody was covered by health insurance so we were, there was too much of antibiotics used so we were educating some of them, went to use antibiotics and some of them didn't and we compared their prescription habits. And they changed after education so that was actually what you could call German cluster trials probably. We didn't call it that way and we didn't look at it as the need to do any consent of people who were given prescriptions. Maybe we were wrong but basically, I think people are better off without too many antibiotics but I mean, that was the implication of the knowledge because if you go too far, then whatever we educate people in medical school should go through consent form or graduate courses? Anyway, so that's my question. Jeremy Grishaw: I think the issues about are you doing the study to generate new knowledge or are you doing a program which is not about generating knowledge, so that's one distinction. So if you're just doing programs without wanting to generate new knowledge or only to match the program results, then really, that takes you out of the research environment but I think what you'll get later in the morning is you'll also get a discussion about when things like waive of consent might be appropriate and they certainly take into account some issues like unlikely risk that potential subjects may occur is part of that process. So I think it would be interesting to reflect on the next presentation whether that addresses power to Karmela. Karmela Krleza-Jeric: Even if you applied, it's always generating new knowledge. We know the nets helps malaria but we wanted to see what. Or we know smoking cessation, we know smoking is no good for your health so it's always new knowledge but it's not necessarily risky so there's the second plot right? Jeremy Grimshaw: But you're still doing research and so the subjects who are in research are basically entitled to having ethical protections and so as soon as you say you're generating knowledge, whether it is social science knowledge about how do we change behaviors, whether it's sort of practical implication about what happens if we change co-payments or whatever. Iif you are doing that for research purposes put you into the research bucket and then we need to sort of think that people are research subjects and then think about one of the appropriate protections. So I think, once you make that distinction, if you're in the research bucket, we have to make a decision then about the benefits and harms and what are the appropriate consent or waive of consent is appropriate in that setting.

Question/Comment from the audience: Peter Tugwell. Over the weekend actually, I had a few manuscripts around the Zelen approach to allocation and I noticed in the, on the sides that most of them were, the correct answer was to do collect the information before the intervention started and before allocation. So I would be interested in your views and whether you will be addressing it over this next few days in terms of either the Zelen, the design is acceptable and what the ethical guidelines will be for that. There are studies that came out of Holland that were working on that and ran into some major ethical problems. Allan Donner: Well, it's very interesting that the Zelen design is proved very controversial from both a methodologies and an ethical perspective in the context of individually randomized trials in cancer. But for years, the Zelen design has been implicitly used in cluster randomization trials without anybody objecting or perhaps even noticing, that is, people are randomized to either experimental group or a controlled group of family practices. The control practices may not even be aware they are in the study and the experimental practice patients only are applied informed consent after they've been randomized. And that of course is the essence of the Zelen design where informed consent is obtained not before but after randomization so to me, it's just interesting that what has been very controversial in individually randomized trials via the Zelen design has been regarded as routine in cluster randomization trials for better or for worse.

Question/Comment from the audience:
Nicholas Birkett: Nick Birkett. I'm also the University of Ottawa. I've been reading, working my way through a book by Margaret Summerville recently which is talking about individualism and collectivism and my question relates to that but, will those sort of issues come into play? It seems that ethics boards are very much focused on individual rights and individual decision making and to a large extent, ignore the collect benefits and collective rights. I guess the question that really comes up though is in the cluster situation. What do you do if, cause groups and clusters are going to heterogenic and as you move from smaller to larger clusters, you get more heterogeneity, politically we're never unanimous at anything and this perfectly, many perfectly legitimate situations where say the example that was given, some people may legitimately say, I don't want to consent to have fluoride in their water and some people might want it in their water. How do you combine those conflicting but perfectly legitimate perspectives into a consent process for a cluster trial? Allan Donner: Well, I think actually this will be addressed in some detail throughout the conference but it's true that you have interventions like water fluoridation for example, where it's very difficult, maybe impossible for someone to escape the intervention, so other speakers will talk about that in much more detail. Jeremy Grimshaw: Not the precise comment you made, but the other issue we're coming to is about clinical equipoise. That individual patients in randomized trials as often as the individual relationship between the clinician and the patient where as often in cluster trials, you’re trying to generate different types of knowledge, you have value to different people. So I think one of the things you're trying to think about is who can make those, what others can see the appropriate judgments where you may be looking at the basically cost effectiveness of say a translation intervention and is really at the health care system level, the knowledge is going to have value but not at the particularly individual level. So I think that's also another area that will come into.

Question/Comment from the audience:
Jonathan Kimmelman: It's really a basic question which is how common is it in the practice of cluster random trials for there to be data safety monitoring committees? Allan Donner: I think that major cluster randomization trials now routinely form independent data monitoring and safety committees. I'm on several myself. Jeremy Grimshaw: I think it depends on what the cluster trial is, so I never have had a data monitoring committee because I'm not sure. I think some of the issues are what the balance of the harms are and also actually detecting proper benefits signal is a real issue so I think there's a violability that might be particularly what the perceived benefits and risks of what you expect is part of this. So when I'm randomizing what's going different quality translator, different quality improvement, the likelihood of harm is very small. We have no idea what the likelihood of benefit but we have...Jonathan Kimmelman: But the data quality and the cluster, for example, could raise concerns about whether you're going to reach out power so there's these parallel issues even if they are around safety issues, correct? Jeremy Grimshaw: Well, often in cluster trials, the ones I do, you know all the cluster crew since day one. It's not sequential. So you may have issues around identifying patients within clusters but normally for the number of clusters, you're actually know that you have 90 family practices that you get accrued. It's not like patient trials. Allan Donner: I think it also depends upon the nature of the intervention and the nature of the outcome. I'm on a data monitoring committee now involved in administrating vitamin aide to neonates and looking at 6 month infant mortality. Well there, I think it would be almost irresponsible not to set up a data monitoring and safety committee. One issue that arises though is that stopping rules for individually randomized trials are well established in the literature but not so well established for cluster trials but from an ethical point of view, I think there's no reason to avoid data monitoring committees in cluster trials and individual randomized trials.